Association between frailty status and delayed observation-room retention after painless gastrointestinal endoscopy...

Frailty reflects reduced physiological reserve and may be associated with slower post-anesthesia recovery in older patients. This study examined the association between frailty status and delayed observation-room retention after painless gastrointestinal endoscopy and explored its additional discriminative contribution beyond routinely collected clinical and peri-procedural variables.

A total of 516 older patients undergoing painless gastrointestinal endoscopy were included, and all analyses reported in the abstract and main manuscript were based on this same cohort. Frailty status was assessed using the Fried frailty phenotype. The primary outcome was delayed observation-room retention, defined using the 75th percentile of observation-room retention time in the study population. Logistic regression models were used to evaluate the association between frailty status and delayed observation-room retention. Model 1 was unadjusted, Model 2 was the primary directed acyclic graph (DAG)-informed confounder-adjusted model, and Model 3 additionally included total propofol dose and intraoperative adverse events as peri-procedural factors. Receiver operating characteristic (ROC) curves, net reclassification improvement (NRI), integrated discrimination improvement (IDI), and decision curve analysis were used for an exploratory in-sample assessment of the additional discrimination provided by frailty status. Because the base model included information arising during the procedure, this analysis was not intended to establish a pre-procedural prediction tool. Sensitivity analyses examined observation-room retention time as a continuous outcome and used the 80th percentile threshold.

The 75th percentile of observation-room retention time was 37.00 min, and 151 patients were classified as having delayed observation-room retention. Compared with patients with normal retention, those with delayed retention had a higher proportion of frailty (72.8% vs. 33.2%, p < 0.001). Frailty was associated with delayed retention in the unadjusted model (OR 5.410, 95% CI: 3.556-8.231, p < 0.001) and in the primary DAG-adjusted model (OR 3.098, 95% CI: 1.490-6.442, p = 0.002). After additional adjustment for peri-procedural factors, the association was attenuated but remained statistically significant (OR 2.061, 95% CI: 1.160-3.662, p = 0.014). Frailty status alone had an AUC of 0.698 (95% CI: 0.655-0.742). In the exploratory peri-procedural model, adding frailty increased the AUC from 0.932 to 0.940 (absolute increase, 0.008; DeLong p = 0.044). The total NRI was 0.665 (95% CI: 0.397-0.998, p = 0.001), and the IDI was 0.021 (95% CI: 0.011-0.067, p = 0.001). Decision curve analysis showed higher apparent net benefit across selected threshold ranges.

Frailty status was associated with delayed observation-room retention in older patients undergoing painless gastrointestinal endoscopy. Its additional discriminative contribution beyond a model containing clinical and peri-procedural variables was small in absolute AUC terms and remains exploratory. Frailty assessment may provide useful information for post-procedure observation planning, but prospective multicenter validation is required before clinical implementation.